David Yallup

[David Yallup @DYallup]

@Plinz I don't think that's the implication of what's being said. It's not the case that mathematicians will unionize to stop the machines, rather on an individual level the incentives to share work in progress are gone. If you don't have 10k agents the smart money is closed source

[David Yallup @DYallup]

@chhaviyadav_ One of the more insidious things is that "fraud" often gets baked in due to the benchmark datasets that we mode collapse on. In AI/ML 4 science there's limited incentive to test on frontier problems that actually challenge the methods

[David Yallup @DYallup]

Excited to be at the #ICLR in Singapore for the next week! I'll be presenting some recent work "Nested Slice Sampling" - where we revisit a well established MC method and give it a GPU accelerated boost.

Poster will be presented at the FPI workshop on the Monday!

[Martin Bauer @martinmbauer]

I hope the prime minister is aware that salaries of postdoctoral researchers have reached such a low level in this country that this law would make it impossible for UK scientists to employ international experts

It would be a huge own goal!

[(((Dan Hodges))) @DPJHodges]

Large numbers of people leaving Tory conference tonight because there are no trains tomorrow. Thereby missing the PM’s speech announcing the axing of the Manchester rail upgrade. Historians will write entire chapters on the events of the past two weeks…

[Dulwich Quantum Computing @DulwichQuantum]

Let's not confuse doing science with doing PR!

Embargoing your research to avoid scrutiny and to maximize the splash when it comes out is a PR tactic. It has nothing to do with critically pursuing the truth through the efforts of a scientific community.

[Mr Demos of Pnyx @gem_ste]

The extreme and endemic nature of rent seeking in the UK, extracting as it does wealth without creating wealth, destroys not only the economy but creates barren towns, cities and societies in which nothing can operate, diversify or develop https://t.co/y6bi5VYVdh

[a lion @paleoludic]

funny that we produced exactly one generation of people who can understand computers

[@emilymbender.bsky.social @emilymbender]

I see people asking: How else will we critically study GPT-4 etc then?

Don't. Opt out. Study something else.

GPT-4 should be assumed to be toxic trash until and unless #OpenAI is *open* about its training data, model architecture, etc.

[Meredith Whittaker @mer__edith]

This. Google just quietly re-started its work on Maven, creating "AI" drone targeting+surveillance tech for the US military after protest killed the contract 5 years ago. I've heard crickets in response, even as the millionth "sorry GPT isn't a person" oped floats across my TL.

[Machine Learning: Science and Technology @MLSTjournal]

Great new work by @PabloLemosP @ElenaMassara @cosmo_shirley @MilesCranmer @changhoon_hahn @DYallup @muntazirabidi @meickenberg @FlatironCCA @NYUPhysics - 'Robust #simulation-based inference in #cosmology with #Bayesian #neuralnetworks' - https://t.co/eokGoja52N #astrophysics #AI

[Philip Oldfield @SustainableTall]

Oh

Study finds if we have a mass global takeup of autonomous vehicles, the powerful onboard computers needed to run them could generate as many greenhouse gas emissions as *all the data centres in operation today*

https://t.co/3eiHtRCU4e

[Keir Starmer @Keir_Starmer]

After 12 years of Tory failure, the British people deserve so much better than this revolving door of chaos.

We need a general election, now.

My full statement:

[darryl li @dcli]

saw ted chiang give a talk yesterday where he basically told a bunch of ai bros, "we are nowhere near having real ai, and what we call ai today is just a tool of capitalism. i don't fear ai, i fear capitalism" and the ai bros got their feelings hurt. it was great.